Shift constraint angle off hor. and vert.

[KeesW]

I have and issue with using the 'SHIFT' constraint to place elements vertically or horizontally. Sometimes, the constraint puts the object just off vert. or hor..The off angle is around 1 degree or even less and it is often too close to notice. It is only when one aligns it with something that is correctly hor. or vert. that it is noticed. Then, it seems to fix itself, until it re-occurs - usually unwittingly. I've had this happen in many recent versions including AC19.

[Erwin Edel]

DGSketcher wrote:
Laszlo, not sure if this is related but I have a simple joist layout formed by 150x50 section beams, if I click on the key points on plan whilst creating a linear dimension line I get circular anchors in usually the right place, however, if I create the dimension line based on the two outer points only and then try to cmd-click to add intermediate points once the tick mark shows I can end up with random dimensions. If I zoom in closer than I should need to then I can generally get the right point but zoomed out with still obvious points I get odd results.

(Recently updated to El Capitan!)

I am having similar issues on windows 7 with ArchiCAD19 lately.

I'm also having issues with edge offset of fills, slabs, roofs etc picking a random point away from the edge (which shouldn't even be possible). I've reported these to the dutch reseller (Kubus). It's hard to reproduce, but it happens multiple times a day.

I've been working on 2d details a lot lately. So lots of work with fills and lots of dimensioning, it becomes very obvious. I'm no novice user, I've been drafting with ArchiCAD for almost 10 years. These problems are very recent. I'd say latest update, but I'm not 100% certain of that.

[DGSketcher]

Laszlo, I am working the same way as I have for the last 15 versions of AC. As Erwin has also mentioned, v19 seems to have brought an element of "randomness" to the snap points affecting various operations which may also be affecting the orthogonal alignment. When I have a lot of lines on screen the blue guidelines appearing can be overwhelming as AC tries to make sense of them all to arrive at an assumed intersection/alignment. Personally I'm beginning to think the new snap guides are not fit for purpose.

It may be worth considering that AutoC*D has an option to limit the type of snaps being found, perhaps AC should be considering something similar.

Just to add a little pressure for GS and a heads up to other users, some of the displacements are quite subtle but if the errors aren't picked up in the design this may lead to some very red faces on site!

[Laszlo Nagy]

Could either of you guys create a short video clip showing a specific issue? So we can show that to the developers?

Regarding Snap References: I think it is best to have only a few of them displayed at any time, because, as you said, when there are too many, there can be a total confusion of Snap Guides.
When there are too many, I just press ESC and all of the disappear. Then I just display the 1 or 2 Snap References I need.

[DGSketcher]

I don't know how to do a video but I have attached a screen grab showing the sort of results that occur when drawing a line between beams. The snap point has no relevance to what I want to draw, it is I think a perpendicular offset to the beam at the top. Just moving the end of the line around generates all sorts of unexpected and potentially random snap points when all I really want is a perpendicular connection. I'm not saying the guides don't work, just that they are counter intuitive and if you aren't zoomed in closely then the snapped point may be off from what is really wanted.

As I said previously I think the snap guides are trying to be too clever!

[Erwin Edel]

Below is a screenshot I sent to the local reseller.

[Laszlo Nagy]

DGSketcher wrote:
I don't know how to do a video but I have attached a screen grab showing the sort of results that occur when drawing a line between beams. The snap point has no relevance to what I want to draw, it is I think a perpendicular offset to the beam at the top. Just moving the end of the line around generates all sorts of unexpected and potentially random snap points when all I really want is a perpendicular connection. I'm not saying the guides don't work, just that they are counter intuitive and if you aren't zoomed in closely then the snapped point may be off from what is really wanted.

As I said previously I think the snap guides are trying to be too clever!

A Snap References Point is created automatically after your cursor finds any node and stays over it for a little time, normally 1 second.
This amount of time can be specified by you in the "Highlight as Snap Reference after" field of the Work Environment Dialog's Input Constraints and Guides page.
If you set it to a large enough value, like 10 seconds or 20 seconds, then Snap Reference Points will practically never appear since you almost never leave your cursor above an element node that long.
You can use this trick along with the Q key on your keyboard, which will actually force a Snap Reference Point (or Line) to appear immediately.
Using these two things you can very well control when and where Snap Reference Points and Lines do and do not appear.
I would recommend watching the ARCHICAD 19 New Features - Guide Lines playlist, it will give you a thorough understanding of the topic:

https://www.youtube.com/watch?v=fMyZfolzzes&list=PLnXY6vLUwlWXj_1QX5iXbZ0cfscL-VNEX

[Laszlo Nagy]

Erwin wrote:
Below is a screenshot I sent to the local reseller.

Erwin,

That indeed looks like something curious is going on.
Would you mind trying to experiment with the 2D Advanced Redraw options in your Work Environment Dialog to see if any other settings makes a difference in this.
You may also try to update your graphics card driver just to makes sure it is not a phenomenon caused by a driver issue.

[DGSketcher]

I think this perhaps offers the best indication of a problem...

https://www.youtube.com/watch?v=knMw7CLh7a4&index=16&list=PLnXY6vLUwlWXj_1QX5iXbZ0cfscL-VNEX

What I don't think is considered by GS in the real world is that where you have a complex drawing then the resulting intersections aren't always what is anticipated. Where previously I might have expected a simple perpendicular connection the snap guides are offering numerous intersections some of which can be misinterpreted.

I guess I can turn the guides off unless really needed. However I am still getting odd results for the dimensioning mentioned previously. If I dimension by clicking the edge or centreline of the beams I get the right result, however, if I click on the intersection points of two beams the resulting dimension can be random. It is also necessary for a beam to be highlighted before a dimension becomes associated which slows things down.

[Erwin Edel]

I've tried to show the dutch reseller Kubus today with TeamViewer when the bug occured again, sadly that cancelled the offset, so I couldn't show it again. They did inform me that Graphisoft has already been told about this, so no need to report this edge offset thing a second time, Laszlo.

I've also updated the drivers for our Quadro K2000, as you suggested (good idea, actually!), that didn't work.

[DGSketcher]

Laszlo, I'm not sure if this was worth a new thread for the linear dimensioning issue.

If as previous I anchor the end points for my dimension line and cmd-clk to add in-between dimension points I am prone to get random static dimensions. If however I highlight the dimension line and then click again to bring up the pet pallet and select insert/merge dimension point then the cursor momentarily changes to a hammer before converting to a tick when the dimension point is reached. The resulting dimension is consistently accurate and normally as expected associative.

Both processes have been tested at the same zoom e.g. my normal working zoom.

To me this is starting to look like a bug in the dimensioning process.

Obviously the latter workflow is also a lot slower than just cmd-clk on a string of points...